Professional Documents
Culture Documents
10.1190/1.3467825
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Manuscript received by the Editor 30 November 2009; revised manuscript received 5 February 2010; published online 14 September 2010.
1
ConocoPhillips, Houston, Texas, U.S.A. E-mail: douglas.j.foster@conocophillips.com; robert.g.keys@conocophillips.com; f.d.lane@conocophillips.com.
© 2010 Society of Exploration Geophysicists. All rights reserved.
75A3
75A4 Foster et al.
reflectivity 共NI or A兲 and Poisson reflectivity PR. Smith and Gidlow AVO response. Two examples illustrate these principles. The first
共1987兲 combine Aki and Richards’ approximation with the mudrock example demonstrates the use of AVO methods to detect hydrocar-
line 共Castagna et al., 1985兲 to define the fluid factor ⌬F. bon-bearing sands, and the second illustrates the use of AVO as a li-
Castagna and Smith 共1994兲 compare several of these seismic at- thology identifier to distinguish reservoir sands from shale.
tributes using velocity and density measurements from a worldwide We begin with our analysis of how changes in elastic properties
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
collection of brine sands, gas sands, and shales. They find that the affect the seismic AVO response.
AVO product A*B detects low-impedance gas sands of the sort de-
scribed by Ostrander 共1984兲 but is ambiguous when it comes to dis- Effects of elastic-property changes on AVO
tinguishing high-impedance gas sands or gas sands with little im-
pedance contrast from brine sands or shales. On the other hand, the For relatively small angles of incidence, usually less than 30°,
reflectivity difference RP ⳮ RS and 共A Ⳮ B兲 / 2 distinguish gas sands Shuey 共1985兲 shows that the compressional-wave reflection coeffi-
from brine sands, regardless of acoustic impedance. cient R can be approximated by an equation of the form
Using the same data set, Smith and Sutherland 共1996兲 show that
R共 兲 ⳱ A Ⳮ B sin2共 兲. 共1兲
⌬F is also able to distinguish gas sands from brine sands, indepen-
dent of the acoustic impedance of the gas sand. Sands can have high- In equation 1, is the angle of incidence, A is the intercept and B is
er or lower acoustic impedance than surrounding shales, but gas the slope of the reflection coefficient evaluated at zero offset. A deri-
sands generally have a much lower Poisson’s ratio than shales or vation of these results is given in Appendix A.
brine sands. The seismic attributes 共RP ⳮ RS兲, 共A Ⳮ B兲 / 2, ⌬F, and For small perturbations in velocity and density at a reflecting in-
PR tend to highlight contrasts in Poisson’s ratio and thus are robust terface, the intercept and slope can be approximated by
gas-sand indicators. The attributes A, RP, and NI are equivalent indi-
cators of acoustic-impedance contrasts. These can be paired with ⌬VP ⌬
A⳱ Ⳮ 共2兲
gradient B, RS, or PR to interpret lithology and pore fluid 共Foster et 2VP 2
al., 1993; Verm and Hilterman, 1995; Castagna et al., 1998兲.
and
再 冎
Reflectivity attributes describe contrasts in elastic properties at a
welded interface and can be converted to impedance attributes by in- ⌬VP V2 ⌬ ⌬VS
version. For example, RP and RS can be inverted to obtain P-wave B⳱ ⳮ 4 S2 Ⳮ . 共3兲
and S-wave impedance. Goodway et al. 共1997兲 use these impedanc- 2VP VP 2 VS
es to compute the elastic properties and , where and are the In equations 2 and 3, VP, VS, and are the averages of the compres-
Lamé parameters and is density. Connolly 共1999兲 introduces the sional-wave velocity, shear-wave velocity, and density above and
concept of elastic impedance by generalizing the inversion of a nor- below the reflecting interface, respectively; ⌬VP, ⌬VS, and ⌬ are
mal-incidence stack to the case of variable angle of incidence. He the differences in compressional-wave velocity, shear-wave veloci-
shows how, for appropriate assumptions, elastic impedance can be ty, and density between these layers, respectively.
derived from Aki and Richards’ 共1980兲 approximation for the angle- Let ␥ ⳱ VS / VP and ⌬␥ represent the difference in this ratio be-
dependent reflection coefficient. More recently, Masters et al. 共2009兲 tween the layer below and the layer above the reflector. Neglecting
generalize the AVO attributes to incorporate a probabilistic classifi- second-order terms,
cation scheme. This provides a Bayesian-derived estimate of the un-
certainty of the attributes. ⌬␥ ⌬VS ⌬VP
⳱ ⳮ . 共4兲
In this paper, we focus on the AVO attributes intercept A and slope ␥ VS VP
B. Most of the attributes previously discussed can be related to A and
B through the Aki and Richards 共1980兲 approximation. Using A and Substituting equation 4 into 3 and combining with equation 2 shows
B is advantageous because they relate directly to the angle-depen- that
dent seismic data. It is easy to predict the effect that a change in A or
⌬
B will have on a common-depth-point 共CDP兲 gather or angle stacks. B ⳱ 共1 ⳮ 8␥ 2兲A ⳮ 4␥ ⌬␥ Ⳮ 共4␥ 2 ⳮ 1兲 .
Thus, understanding the impact of changes in reservoir properties on 2
A and B provides insight when interpreting the seismic-amplitude
The assumption that VP / VS ⳱ 2.0 is often a good approximation for
response.
normally pressured shale. If the ratio ␥ is close to 0.5 共VP / VS ⳱ 2.0兲,
We begin by analyzing the effect of changes in elastic properties
the last term can be neglected as a second-order perturbation, yield-
on intercept and slope. We show that the key elastic properties which
ing
control the angle-dependent reflection coefficient are acoustic-im-
pedance contrast and contrast in Poisson’s ratio, or, equivalently, B ⳱ 共1 ⳮ 8␥ 2兲A ⳮ 4␥ ⌬␥ . 共5兲
VP / VS. From this analysis, we can explain the AVO behavior ob-
served by Rutherford and Williams. In a crossplot of A versus B, equation 5 describes a family of lines
Our analysis is based on Aki and Richards’共1980兲 approximation. that is approximately parallel to the line:
Because this is a linear approximation that assumes small perturba-
B ⳱ 共1 ⳮ 8␥ 2兲A. 共6兲
tions in elastic properties, we investigate the consequences of the
small contrast assumption. We include a discussion on the effect of We call this the fluid line.
large or nonlinear changes in elastic properties. Equation 5 shows that in the crossplot domain, reflections are
From our analysis on the effects of elastic properties on the AVO characterized by their contrasts in acoustic impedance and VP / VS
response, we determine the effects of changes in reservoir proper- relative to the fluid-line trend. An intercept and gradient crossplot
ties, such as pore fluids, porosity, and clay content, on the seismic based on equation 5 is depicted in Figure 1. AVO responses at the top
Interpretation of AVO anomalies 75A5
of sand are shown for the four classes of gas sands identified by Ru- intercept and a slope that is zero or positive. The reflection from the
therford and Williams 共1989兲 and Castagna and Swan 共1997兲. The top of a class IV gas sand is negative, but its magnitude does not in-
slope of the fluid line depends on the background VP / VS 共or VP / VS crease with angle.
⳱ 1 / ␥ 兲. The fluid-line slope is ⳮ1 if VP / VS ⳱ 2 and rotates counter-
clockwise as VP / VS increases. These observations about the fluid Exact As and Bs
line hold whether the background VP / VS is constant or slowly vary-
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
IV
sand trend with the slope axis. The reflection from the top of a class II
gas sand is negligible at zero offset but has a negative slope, so its III
amplitude becomes large in magnitude with respect to the zero-off-
set amplitude and negative with increasing angle. δVP /VS II
Reducing acoustic impedance further leads to a class III gas sand
that has lower impedance than the overlying shale. Figure 1 shows Top of sand I
that the reflection from the top of a class III gas sand has negative in-
Intercept
tercept and slope; consequently, it is negative at normal incidence
and becomes more negative with increasing angle.
Figure 1. Intercept A versus slope B crossplot. AVO responses at top
Continuing to decrease the acoustic impedance moves the reflec- of sand are shown for the four classes of gas sands. The polarity con-
tion intercept-slope point up and to the left on the gas sand trend to vention for this plot denotes a decrease in acoustic impedance by a
produce a class IV gas-sand reflection, characterized by a negative negative amplitude 共trough兲.
75A6 Foster et al.
means the linear relationship between slope and intercept is most ac- nate that base-of-sand reflections have this enhanced AVO response
curate when VP / VS is near 2.0. When VP / VS ⬎ 2.0, there is more because they can be used to identify downdip limits that provide fur-
scatter from the trend that is unrelated to changes in VP / VS.Addition- ther support for the presence of hydrocarbons. This enhanced re-
ally, the density term that was omitted in equation 5 becomes more sponse can be seen in the B attribute and any other attributes that are
significant. This is evident in Figure 2a, where the slope and inter- a function of the slope 共e.g., fluid line, Poisson reflectivity, fluid fac-
tor, A Ⳮ B兲. As a second-order effect, this asymmetry may not be vis-
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
cept are strongly correlated when VP / VS ⳱ 2.0 but are less correlated
when VP / VS is different from 2.0. Also, for larger magnitudes of A, ible on a conventional seismic section for thin sands. In practice, the
the background trend is not linear. best opportunity to observe the effect is with relatively thick reser-
These results suggest that AVO methods which use distance from voirs bounded by a uniform shale that produces no interference be-
a background trend to detect hydrocarbons are often more effective tween the top and base reflections.
in sediments where the background VP / VS is close to 2.0. In shallow,
unconsolidated sediments where VP / VS ⬎ 2.0, the background
Effects of rock- and fluid-property changes on AVO
trend will be less correlated. This can also be true in overpressured
sediments where the background VP / VS can exceed 2.0 at great Up to this point, we have considered the effects of changes in elas-
depths. Intercept or normal-incidence reflectivity can be a better hy- tic properties such as acoustic impedance and VP / VS on the seismic
drocarbon indicator in shallow, unconsolidated sands than an AVO AVO response. The more important issue is the effect of changes in
anomaly. Note that class III or IV gas sands may produce a large in- rock and fluid properties on the response. One property that has a sig-
tercept or normal-incidence reflection coefficient. Polarity is impor- nificant effect is pore-fluid compressibility.
tant in this case. Replacing brine with a highly compressible pore fluid such as gas
The second difference between equations 5 and 7 is the perturba- or light oil reduces the compressional-wave velocity of the rock. Al-
tion term containing ⌬␥ . Equation 7 implies that trends resulting though the shear modulus is unaffected by the type of pore fluid, the
from changes in ␥ are not symmetric with respect to the fluid line. If shear-wave velocity increases slightly because of the lower density
all other factors are equal, base-of-sand reflections lie farther from of hydrocarbons. Consequently, increasing pore-fluid compressibil-
the fluid-line trend than top-of-sand reflections. Although symmet- ity significantly reduces the VP / VS of the rock. Equations 5 and 7
ric with respect to normal-incidence reflectivity, equation 7 predicts show that an abrupt change in VP / VS displaces the AVO response
that the AVO slope B response from the base of sand should be more from the fluid-line trend by an amount dependent on the VP / VS con-
prominent than the slope from the top of sand. This asymmetry is a trast. The magnitude of the displacement from the fluid line increas-
second-order effect, but it is evident in Figure 2b. The asymmetry es as pore-fluid compressibility increases.
becomes greater as the VP / VS contrast increases. Actually, it is fortu- The effect of pore-fluid compressibility on AVO responses is de-
picted in Figure 3. This effect is similar to the results obtained by
a) Batzle et al. 共1995兲 for various pore fluids, including 20° API oil,
0.2
VP /VS: 1.5
50°API live oil, and gas. Gas, with the highest compressibility, pro-
duces the greatest departure from the fluid-line trend, followed by
0.1 VP /VS: 2.0 50°API live oil. Heavier oils with low gas content approach the re-
sponse of brine-saturated sands.
Slope
0.0 VP /VS: 2.5 Porosity is another rock property that has a significant effect on
seismic response. An increase in porosity decreases compressional-
Predicted
–0.1
trends wave velocity and density. Unlike fluid compressibility, which has
little effect on shear-wave velocity, an increase in porosity also de-
–0.2 creases shear-wave velocity. The decrease in shear-wave velocity
–0.2 –0.1 0.0 0.1 0.2 offsets the decrease in compressional-wave velocity so that VP / VS is
Intercept not significantly changed. Brie et al. 共1995兲, for example, report a
VP / VS of 1.58 for clean gas sands, irrespective of porosity.
b) 0.6 The effect of porosity changes on the AVO response are illustrated
in Figure 3. Because increasing the porosity of a gas sand reduces its
0.3 acoustic impedance, the intercept A of a reflection from the top of the
sand becomes more negative and moves to the left in Figure 3. How-
Slope
0.0 ever, because porosity changes do not affect the VP / VS contrast, the
slope-intercept value of the reflection remains on a trend defined by
–0.3 the initial VP / VS contrast.
To illustrate how porosity affects the AVO response of a seismic
–0.6 reflection, suppose we observe a reflection from the top of a class III
–0.2 –0.1 0.0 0.1 0.2 gas sand designated by point 1 in Figure 3. At normal incidence, the
Intercept reflection from this sand is negative and becomes more negative
with increasing offset. If we increase the porosity of this gas sand, its
Figure 2. Comparison of predicted trends from equation 7 with slope AVO response will move in the direction of the arrow denoting in-
and intercept values calculated using sonic and density logs from creasing porosity in Figure 3 to point 2. At this new location, the re-
well A 共Keys and Foster, 1998兲. 共a兲 Shear-wave velocities derived
using VP / VS ⳱ 1.5 共red兲, 2.0 共green兲, and 2.5 共blue兲. 共b兲 Shear-wave flection is larger in magnitude 共more negative兲 but has less variation
velocities derived using VP / VS ⳱ 1.9 for background rocks and 1.5 with offset than the reflection at point 1. Alternatively, if we reduce
for sands. the porosity of the gas sand, we will move toward point 3 on the
Interpretation of AVO anomalies 75A7
crossplot. The resulting reflection will have a small amplitude at nor- depend on the pure-sand and pure-shale porosities. As Figure 3 indi-
mal incidence but will increase in magnitude 共become more nega- cates, acoustic impedance can be an ambiguous discriminator be-
tive兲 with increasing angle of incidence. The amplitude increase tween sands and shales, but sands generally have greater VP / VS con-
with angle is greater at point 3 than at point 1 or 2. The reflection at trast with the fluid line than shales.
point 4 results from replacing gas with brine. Reducing the porosity The effects of pore-fluid changes in the A and B crossplot domain
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
of the brine-saturated sand will move the reflection from 4 to 5 on the can be illustrated using the well-log data displayed in Figure 4. Fig-
crossplot. The reflection at point 5 will be large and positive at zero ure 5 shows the crossplot for gas and brine against the shale back-
offset, and its amplitude will decrease in magnitude with increasing ground. The brine points are calculated by Gassmann fluid substitu-
angle. tion 共Mavko et al., 1998兲. In this plot, shale-on-shale reflections de-
In addition to porosity and pore fluid, shale content affects the fine the fluid line. Gas-sand reflections are the farthest from the fluid
seismic AVO response. However, the impact of shale content on line, and brine sands are between the gas-sand and shale points.
AVO characteristics is complex. Figure 3 depicts the AVO effects of Using the well-log data in Figure 4, the effects of porosity changes
increasing shale content based on the sand-clay mixture model de- in the A and B crossplot domain are shown in Figure 6. The porosity
scribed by Marion et al. 共1992兲. Consider the wet-sand point 4 in trend is clearly seen in this plot. The highest-porosity sand has a class
Figure 3. Increasing the shale content by adding clay to the pore IV AVO response, and the lowest-porosity sandstones are class I to
space increases acoustic impedance by reducing porosity. Also, in- class II. This porosity trend is commonly seen in normally compact-
creasing shale content increases VP / VS because the slope and inter- ing sediments; AVO anomalies will follow this porosity/depth trend.
cept response of pure shale must lie on the fluid line. Therefore, in- Overpressured sediments can disrupt the trend. In normal-pressured
creasing shale content moves the AVO response in the direction of sediments, it is unusual to see class III anomalies at significant
the solid brown arrow, which is in the direction of reduced porosity depths below class II anomalies.
but closer to the fluid line.
Clay can be added to the pore space until critical concentration is Example 1: Application of AVO analysis to hydrocarbon
reached. At this point, the pore space is filled with clay, and addition- detection
al clay cannot be added without displacing grains of sand. Increasing
Our first example illustrates the use of AVO analysis to differenti-
shale content beyond the critical concentration reduces acoustic im-
ate light hydrocarbon 共oil and gas兲 sands from brine sands in a struc-
pedance and moves the AVO response in the direction of the dashed
tural trap with four-way dip closure. The study area is outlined in
brown arrow. The start and end points of the shale-content arrows
Figure 7, which shows a time structure map of the top of the reservoir
Fluid line
interval. The structure consists of an eastern, central, and western
Base of sand
substructure. The reservoir sands are stacked and consist of relative-
ly thin fluvial, marine, and deltaic sands with interbedded shales. A
stratigraphic column is shown in Figure 8. Two wells from the study
area that encountered gas, oil, and brine sands are used forAVO anal-
Fluid
Porosity ysis. Well 1 was drilled in the central structure, and well 2 is located
compressibility
in the western structure. Figure 9 shows a far-angle stack section
through well 1 in the central structure.
Slope
1650
tion response farther from the fluid-line trend 共not necessarily in a di-
rection perpendicular to the trend, however兲. An increase in porosity
moves the reflection response parallel to the fluid-line trend, in the
direction of the solid arrows. The numbered points on the crossplot
illustrate the effect of varying porosity and pore fluid on the AVO re-
sponse from top of sand: 1 — AVO response from top of a class III
1700
gas sand; 2 — AVO response of a higher-porosity gas sand; 3 —
AVO response of a lower-porosity gas sand; 4 — wet-sand response
obtained by replacing gas with brine; 5 — AVO response of a lower- Figure 4. Log data from well 1. Track 1 — shale volume curve
porosity wet sand. The solid brown arrow depicts the effect of in- 共green兲. Red curves, tracks 2–5: measured P-wave velocity, S-wave
creasing shale content in a shaly sand by adding clay to the pore velocity, bulk density, and VP / VS. Green curves, tracks 2–5: shale
space until critical concentration is reached, at which point no more trend curves for P-wave velocity, S-wave velocity, density, and
clay can be added to the pore space without displacing grains of VP / VS. Blue curves, tracks 2–5: P-wave velocity, S-wave velocity,
sand. The dashed brown arrow shows the effect of increasing shale density, and VP / VS obtained from the measured 共red兲 curves by re-
content in excess of the critical concentration. placing in situ hydrocarbons with brine.
75A8 Foster et al.
lower interval as a normal consequence of compaction and variation the extent of hydrocarbons. There is no attempt to distinguish gas
in rock quality, resulting from changes in depositional environment. from oil because the oil is light with a high gas/oil ratio.
The aim of this study is to evaluate the potential for differentiating The first step in our analysis is to determine the expected AVO re-
hydrocarbons from water using AVO information and to determine sponse for brine- and hydrocarbon-filled sands. We use the available
well-log data to accomplish this objective. Intercept and slope cross-
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
plots derived from the two wells are shown in Figure 10. Figure 10a
0.5 and b plots are from well 1; Figure 10c and d plots are from well 2 for
0.4 Fl
the same formation. Figure 10a and c shows intercept and slope
ui crossplots for reflections from top of sand and top of shale relative to
d
0.3 lin
e the average background shale for sands and shales in the upper hy-
0.2
0.1 NWa
Slope
NWb
7 km
0.0 N
9 km
–0.1
–0.2
8 km
6 km
–0.3
Central
–0.4 West 12 km
–0.5
–0.5 –0.4 –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 0.5
12 km
Intercept
East
10 km
Figure 5. Slope-intercept crossplot from well 1. Red points are
slope-intercept values of reflections from the background shale/sand
interface for sands from well 1 between 1600 and 1692 m. Blue Figure 7. Time structure map from the top of the productive zone.
points correspond to reflections from the background shale/sand in- The main structure consists of three individual structures 共east, cen-
terface for fluid-substituted brine sands between 1600 and 1692 m. tral, and west兲. Warmer colors 共orange, yellow兲 represent structural
Brown points denote reflections from background shale/shale inter- highs; cooler colors 共purple, blue兲 are structural lows.
face for shales between 1600 and 1692 m. The solid green line is the
fluid line from equation 6 for the background shale VP / VS. The ma- Lithology Saturation
genta line is the trend curve corresponding to a decrease in VP / VS
from 2.0 to 1.67. Figure 4 shows that most gas sands have VP / VS
⬍ 1.67 and, as predicted by equation 5, lie below the magenta trend
line.
Top gas sand
Porosity
0.5 3.6E–01
1700 m
0.4 3.2E–01
Middle Jurassic
0.3
d
2.8E–01
lin
e
0.2
2.4E–01
0.1
Slope (B)
2.0E–01
1800 m
0.0
1.6E–01
–0.1
In oro
1.2E–01
cr sit
p
ea y
–0.2
sin
8.0E–02
g
–0.3
4.0E–02
1900 m
–0.4
–0.5 0.0E+00
–0.5 –0.4 –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 0.5
Intercept (A)
drocarbon-filled reservoir interval. Also shown are the intercept and From 3D prestack-time-migrated gathers, we estimate intercept
slope values for reflections from top of brine sands relative to back- and gradient volumes. Figure 11 shows a crossplot derived from the
ground shale. Brine-sand properties are obtained by substituting the 3D seismic data. The blue points are background data from a region
in situ pore fluid with brine. Intercept and slope values are calculated downdip from the crest of the structure in a section containing wet
from equations 2 and 3, using background shale properties for the sands and shales. The seismic response from this location is used to
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
overlying shale layer and the well-log data for the underlying sand or define the fluid line. The red points come from an updip region on the
shale layer. crest of the structure and in the upper portion of the reservoir inter-
Figure 10b and d shows slope and intercept values derived from val. This part of the reservoir contains oil and gas. TheAVO response
synthetic seismic gathers modeled from the well-log data. Intercept in this upper portion of the reservoir is predominantly class III, as ex-
and slope are determined by a least-squares fit to the angle-depen-
dent reflection coefficient. The red points are slope and intercept val- a) c)
0.50 0.50
ues from the hydrocarbon sand interval, and the blue points are from Top
Top
background or nonhydrocarbon-bearing reflections. Because the 0.25 0.25
lower crossplots are derived from synthetic seismic data, they are af-
Slope
0.00 0.00
fected by the seismic bandwidth. Thus, Figure 10a and c indicates
Base In Situ In Situ
the expected AVO response for individual sand units, and Figure 10b –0.25
Brine
–0.25 Base
Brine
and d shows the impact of thickness and tuning. Shale Shale
–0.50 –0.50
The crossplots and seismic data for this example use the polarity –0.5 –0.25 0 0.25 0.5 –0.5 –0.25 0 0.25 0.5
convention that denotes a decrease in acoustic impedance by a posi- b) d)
tive reflection amplitude, or peak, on a seismic display. Subsequent- Top Top
0.25 0.25
ly, top-of-sand reflections lie above the fluid line and base-of-sand
Slope
reflections lie below the fluid line. –0.25 0.25 –0.25 0.25
Figure 10 illustrates a distinct difference in the AVO response of
–0.25
brine- and hydrocarbon-filled sands observed at the top of the reser- Base –0.25 Base
voir interval. At the top of the reservoir, hydrocarbon sands have a
class III response, approaching class IV.Areflection from the top of a Intercept Intercept
gas sand should be a peak at zero offset and become larger with in-
creasing angle. Figure 10 also shows that brine sands have low im- Figure 10. Slope-intercept crossplots from top gas sand in wells 1 共a,
pedance with less separation in AVO response from background b兲 and 2 共c, d兲. 共a兲 Slope-intercept crossplot using log data from a
shales. This suggests that on a correctly processed seismic section, well in the central structure. The data are from an interval at the top
of the reservoir. 共b兲 Intercept and slope values derived from synthet-
amplitudes should be observed to dim downdip from a hydrocarbon/ ic seismic data from the same interval. 共c, d兲 Slope-intercept cross-
water contact. plots are similar displays from a well in the western structure.
Similar analysis applied to deeper intervals in wells 1 and 2 shows
–60 –48 –36 –24 –12 0 12 24 36 48 60
that the expected AVO response for hydrocarbon sands ranges from
60 60
class III at the top of the reservoir section to class II in the deeper sec-
tion as porosity decreases. Although the brine sands are low imped-
48 48
ance at the top of the reservoir, in general, the acoustic impedance
and VP / VS contrast between brine sands and background shales are 36 36
small, and differences in pore fluids should be evident in the seismic Top
data. With these expectations from the analysis of the well data, we 24 24
proceed to analyze the seismic data.
12 12
Time (ms)
Slope
0 0
1100 Far-angle stack
–12 –12
1200
–24 –24
1300 Base
–36 –36
pected. Using the concepts outlined earlier, we interpret the anoma- ing AVO anomalies is shown in Figure 13a. The AVO color-class
lies in terms of changes in pore fluids and relative-porosity varia- section is overlain on the near-angle 共0°–15°兲 stack, displayed as
tions. wiggle traces. A zero-offset synthetic seismic trace is posted at the
Figure 12 depicts a color scheme used to classify seismic anoma- well 1 location. The AVO color section has dark-green over light-
lies based on Rutherford and Williams’ 共1989兲 classification. Using green amplitude at the crest of the structure between 1200 and
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
this color scheme, a seismic cross section through well 1 highlight- 1250 ms, indicating a class III top over class III base — consistent
with high-porosity gas sands. A notable characteristic of the section
is that the seismic class III anomaly’s downdip termination closely
matches the time-converted hydrocarbon/water contact 共HWC兲
Class I Class II Class III from a multidynamic test 共MDT兲. Throughout the section, the
Class IV strength of the anomalies decreases with depth, and class II anoma-
To
po
fs
lies are seen at deeper depths, consistent with reduced-porosity
an
d sands.
Gas An AVO color-class section through well 2 is shown in Figure
Slope
Oil
Class IV
13b, overlain with the near-angle stack wiggle-trace cross section
Oil
Ba
se and a zero-offset synthetic trace posted at the well location. Like the
of
sa
nd
Gas
cross section through well 1, there is a class III top and base 共dark
green/light green兲 AVO anomaly at the crest of the western structure.
Class III Class II Class I The AVO anomaly terminates downdip at the HWC 共1243 ms兲 es-
Intercept tablished from the MDT.
Map-view amplitude extractions are useful for checking con-
Figure 12. AVO classification scheme for identifying the magnitude formance to structure. Figure 14 shows a horizon amplitude extrac-
and class of a seismic reflection. The polarity convention in this dis- tion from the top of a hydrocarbon-bearing sand using the AVO clas-
play denotes a decrease in acoustic impedance by a peak. sification scheme described previously, compared with an amplitude
extraction from the far-angle stack. The far-angle 共20°–35°兲 stack,
like the fluid-line attribute, is sensitive to VP / VS 共⌬␥ 兲 contrasts.Also
a) shown is a constant time contour corresponding to the HWC estimat-
Time (ms) Well 1 ed from the two wells. This anomaly shows relatively good conform-
1100 ance to structure. The amplitude extraction from the AVO class
GWC 1230 ms volume and the far-angle stack shows the anomaly extends to the
1200
eastern structure within the expected contour interval. A well in the
1300 eastern structure confirms the presence of hydrocarbons in this inter-
val.
1400 This analysis suggests that the seismic AVO response can distin-
guish hydrocarbon-bearing sands from brine sands and shales in the
1500
study area. Consequently, the seismic AVO data were used to deter-
1600 mine the extent of hydrocarbons and to aid in the delineation of the
field.
23.5 km
b)
Slope
1400
1500 10 km
Intercept
Far-stack
1600
Polarity
Well 1
1700
x sdsmlc Well 2 Well 3
29 km
Example 2: Application of AVO analysis to lithology As a general rule, class II anomalies are associated with sands of
discrimination lower porosity than sands that produce class III anomalies.An exam-
ple of this is shown in Figures 15 and 16. This example assumes the
A second example demonstrates the use of AVO analysis for li- polarity convention that a peak denotes an increase in acoustic im-
thology prediction. In hard-rock or low-porosity reservoirs, the fluid pedance. Figure 15a shows the seismic gathers and a crossplot of the
effect on an AVO response can be negligible, but finding reservoir-
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
intercept and slope derived from data around well A. The crossplot
quality sand is important. In this case, fluid contacts and conform- and gathers indicate a class III response for the reservoir. Evidence
ance of seismic amplitude to depth structure may be difficult to dis- of this is seen at the near offset, where there is a trough-and-peak
cern; however, calibrated anomalies consistent with a working geo- pair. The porosity in this well is high enough that the reservoir is con-
logic model may be used to help map sand distribution. The reservoir sidered to be commercial. Figure 15b shows data from an area
in this example is a stratigraphic trap containing light hydrocarbons. around well B. At this location, there is very little energy on the near
Stratigraphically, we expect to see thick sands in a main channel offsets and significant energy at the far offsets. The AVO response is
feeding a turbidite fan. Furthermore, porosities are expected to de- class II. Here, the porosity is lower than that encountered in well A.
crease in a direction distal to the sediment source. The reservoir porosity at well B is below the commercial threshold.
Figure 16 shows the AVO class response in map view. The thresh-
a) old used to create this map highlights the cleanest sands 共farthest
Time (ms) from the fluid-line trend兲. This extraction is taken from the reflection
1400 at the top of the reservoir interval. In Figure 16, red and yellow colors
2.5
1500 Base of represent class III reflections, and blue colors denote class II events.
Reservoir reservoir
Well A penetrated the reservoir, which has a strong class III anomaly,
1600
and found relatively clean, hydrocarbon-bearing sand with good po-
Slope
1700 rosity. Well B, however, penetrated a class II anomaly and found hy-
0.0
1800 drocarbon sand with poorer porosity. This plot identifies the pre-
1900
Top of
reservoir
ferred porosity zones based on the theoretical model described
above. The results discussed here were used in the exploration and
–2.5
2000
–2.5 0.0 2.5
appraisal phases of this area.
2100 Intercept
b) CONCLUSIONS
Time (ms)
Exact expressions for intercept and slope show that the fluid-line
2.5
1400
Base of trend has the least scatter when VP / VS ⳱ 2. In this case, density con-
1500 Reservoir reservoir
trasts do not contribute to the scatter of points about the trend. Analy-
1600 sis of second-order effects shows the reflection-coefficient slope is
Slope
1700
1800 VP / VS. Consequently, all other factors being equal, the gradient of
Top of
1900
reservoir the reflection coefficient from the base of sand is more prominent
than the top-of-sand reflection-coefficient gradient. This knowledge
–2.5
2000
–2.5 0.0 2.5 may aid in determining the downdip extent of the anomaly. Slope
2100 Intercept and intercept crossplots are useful for interpreting AVO anomalies
and explaining the effects of changes in rock and pore-fluid proper-
Figure 15. Gathers around 共a兲 well A and 共b兲 well B locations and the ties. As pore-fluid compressibility increases, the slope and intercept
crossplots derived from the data. These data in the crossplot are from points move away from the fluid line. This can aid in discriminating
the reservoir interval. At the well A location 共a兲, there is a class III
anomaly, indicating relatively high porosity. At the well B location hydrocarbon from brine sands. Gas causes the largest deviation from
共b兲, there is a class II anomaly, indicating relatively low porosity. the trend, and brine has the least; oil-sand points lie in the region be-
tween, based on specific oil properties.
Typically, there is a contrast in VP / VS between clean sand and
N shale. This contrast produces a deviation in the seismic AVO re-
sponse of clean sands from the background trend. Porosity varia-
tions affect acoustic impedance but do not significantly affect VP / VS;
therefore, porosity variations move points approximately parallel to
Well B
Well A the fluid line. These observations can be used to discriminate reser-
I II
voir quality and to distinguish sand from shale. AVO classes can in-
IV III dicate relative changes in porosity. In a relative sense, the highest-
III IV
II
porosity gas sands result in a class IV event, and the lowest-porosity
10 km I
gas sands are characterized by class I reflections. It is important that
any AVO anomaly be interpreted within the context of an appropri-
Figure 16. Amplitude extraction along the top reservoir reflection
from the AVO classification volume. Well A shows a relatively ate geologic model. The principles described here are useful for
strong class III anomaly, and well B shows a class II anomaly. The gaining insight into the geologic controls responsible for the charac-
porosity in well A is higher than the porosity in well B. ter of AVO anomalies.
75A12 Foster et al.
ACKNOWLEDGMENTS ac ⳮ 1
A⳱ 共A-3兲
We acknowledge Jeff Malloy, Scott Irvine, Erik Keskula, Jon ac Ⳮ 1
Anderson, Duncan Emsley, Tim Wallace, Mike Faust, and Renee and
Hannon for their contributions. In addition, we are grateful to re-
8k关k ⳮ ac共b Ⳮ d兲兴 Ⳮ ac关共c2 ⳮ 1兲共b Ⳮ ad兲 ⳮ 2共1 ⳮ a兲2bcd兴
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
APPENDIX A reflection coefficient A, and equation A-4 is the exact slope B, or de-
rivative of the reflection coefficient with respect to sin2共 兲 at zero
SECOND-ORDER RELATIONSHIP offset.
BETWEEN A AND B The small contrast approximations, given in equations 2 and 3,
are derived by defining ⳱ 共 2 Ⳮ 1兲 / 2, ⌬ ⳱ 共 2 ⳮ 1兲, ␣ ⳱ 共 ␣ 2
The derivation of equations 5 and 7 and the attributes A and B Ⳮ ␣ 1兲 / 2, ⌬␣ ⳱ 共 ␣ 2 ⳮ ␣ 1兲,  ⳱ 共  2 Ⳮ  1兲 / 2, and ⌬ ⳱
共equations 2 and 3兲, linearized with respect to medium property con- 共  2 ⳮ  1兲; then substituting these expressions into equations A-3
trasts, is given in Foster et al. 共1997兲. We summarize the results be- and A-4 and retaining only first-order terms.
low. Assuming the density contrast is negligible,
The coefficients of the reflected R and transmitted T plane waves
1ⳭA
at an interface between two elastic solids are determined by the con- c⳱ , 共A-5兲
ditions that normal and tangential components of stress and dis- 1ⳮA
placement must be continuous. In matrix form 共Achenbach, 1973, p. where a ⳱ 1, b ⳱ ␥ 1, d ⳱ c␥ 2, ␥ 1 ⳱  1 / ␣ 1, and ␥ 2 ⳱  2 / ␣ 2. Sub-
186兲, the four equations are stituting these into equation A-4 yields an expression for B in terms
of A, ␥ , and ⌬␥ . Expanding the denominator of this expression for B
冤冥冤 冥
RPP x with respect to A and collecting terms by powers of A produces equa-
RPS 冑1 ⳮ x2 tion 7.
M ⳱ , 共A-1兲
TPP 2b2x冑1 ⳮ x2
REFERENCES
TPS 1 ⳮ 2b2x2
Achenbach, J. D., 1973, Wave propagation in elastic solids: North-Holland
where M is the matrix Publishing Co.
Aki, K., and P. G. Richards, 1980, Quantitative seismology: Theory and
methods, vol. 1: W. H. Freeman & Co.
冤 冥
ⳮx ⳮ冑1 ⳮ b2x2 cx ⳮ冑1 ⳮ d2x2 Batzle, M. L., D. Han, and J. P. Castagna, 1995, Fluid effects on bright spot
and AVO analysis: 65th Annual International Meeting, SEG, Expanded
冑1 ⳮ x 2 ⳮbx 冑1 ⳮ c 2 x 2 dx Abstracts, 1119–1121.
Brie, A., F. Pampuri, A. F. Marsala, and O. Meazza, 1995, Shear sonic inter-
2b2x冑1 ⳮ x2 b共1 ⳮ 2b x 兲
2 2
2ad2x冑1 ⳮ c2x2 ⳮad共1 ⳮ 2d2x2兲 pretation in gas-bearing sands: 70th Annual Technical Conference, Soci-
ety of Petroleum Engineers, Proceedings, 701–710.
ⳮ共1 ⳮ 2b2x2兲 2b2x冑1 ⳮ b2x2 ac共1 ⳮ 2d2x2兲 2ad2x冑1 ⳮ d2x2 Castagna, J. P., M. L. Batzle, and R. L. Eastwood, 1985, Relationships be-
tween compressional-wave and shear-wave velocities in clastic silicate
共A-2兲 rocks: Geophysics, 50, 571–581.
Castagna, J. P., and S. W. Smith, 1994, Comparison of AVO indicators: A
modeling study: Geophysics, 59, 1849–1855.
and where a ⳱ 2 / 1, b ⳱  1 / ␣ 1, c ⳱ ␣ 2 / ␣ 1, d ⳱  2 / ␣ 1, and x Castagna, J. P., and H. W. Swan, 1997, Principles of AVO crossplotting: The
Leading Edge, 16, 337–342.
⳱ sin . The angle is the angle of incidence, measured counter- Castagna, J. P., H. W. Swan, and D. J. Foster, 1998, Framework for AVO gra-
clockwise from the normal to the reflecting boundary. The parame- dient and intercept interpretation: Geophysics, 63, 948–956.
ters characterizing the properties of the lower half-space ␣ 2,  2, and Connolly, P., 1999, Elastic impedance: The Leading Edge, 18, 438–452.
Domenico, S. N., 1977, Elastic properties of unconsolidated porous sand res-
2 are the compressional velocity, shear velocity, and density, re- ervoirs: Geophysics, 42, 1339–1368.
spectively. The parameters ␣ 1,  1, and 1 are similarly defined for Fatti, J. L., G. C. Smith, P. J. Vail, P. J. Strauss, and P. R. Levitt, 1994, Detec-
tion of gas in sandstone reservoirs using AVO analysis: A 3-D seismic his-
the upper half-space. The compressional and converted shear reflec- tory using the Geostack technique: Geophysics, 59, 1362–1376.
tion coefficients are RPP and RPS, and the compressional and convert- Foster, D. J., R. G. Keys, and D. P. Schmitt, 1997, Detecting subsurface hy-
ed shear transmission coefficients are TPP and TPS. drocarbons with elastic wavefields, in G. Chavent, G. Papanicolaou, P.
Sacks, and W. Symes, eds., Inverse problems in wave propagation: Spring-
Let D denote the determinant of M, and let N be the determinant er-Verlag, 195–218.
of the matrix obtained by replacing the first column of M with the Foster, D. J., S. W. Smith, S. Dey-Sarkar, and H. W. Swan, 1993, A closer
look at hydrocarbon indicators: 63rd Annual International Meeting, SEG,
vector on the right-hand side of equation A-1. The compressional- Expanded Abstracts, 731–734.
wave reflection coefficient is given by RPP ⳱ N / D. Goodway, W. N., T. Chen, and J. Downton, 1997, Improved AVO fluid detec-
The derivation of equation 1 assumes relatively small angles of tion and lithology discrimination using Lamé petrophysical parameters;
“ ,” “ ,” and “ / fluid stack,” from P and S inversions: 67th Annual
incidence. With this assumption, RPP is approximated with a Taylor- International Meeting, SEG, Expanded Abstracts, 183–186.
series expansion with respect to sin2 , 共x2兲 evaluated at ⳱ 0. Gregory, A. R., 1976, Fluid saturation effects on dynamic elastic properties
of sedimentary rocks: Geophysics, 41, 895–921.
From Foster et al. 共1997兲, the expressions for A and B for arbitrarily Keys, R. G., and D. J. Foster, eds., 1998, Comparison of seismic inversion
large contrasts are methods on a single real data set: SEG.
Interpretation of AVO anomalies 75A13
Koefoed, O., 1955, On the effect of Poisson’s ratios of rock strata on the re- nonnormal angles of incidence: Geophysics, 49, 1637–1648.
flection coefficients of plane waves: Geophysical Prospecting, 3, 381– Rutherford, S. R., and R. H. Williams, 1989, Amplitude-versus-offset varia-
387. tions in gas sands: Geophysics, 54, 680–688.
Marion, D., A. Nur, H. Yin, and D. Han, 1992, Compressional velocity and Shuey, R. T., 1985, A simplification of the Zoeppritz equations: Geophysics,
porosity in sand-clay mixtures: Geophysics, 57, 554–563. 50, 609–614.
Masters, A. R., S. Y. R. Gesbert, K. M. Wojcik, C. K. Chan, and L. P. Jong, Smith, G. C., and P. M. Gidlow, 1987, Weighted stacking for rock property
2009, Stalking uncertainty with Bayesian visualization: 71st Conference
estimation and gas detection: Geophysical Prospecting, 35, 993–1014.
Downloaded 04/19/15 to 197.16.156.242. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/